Oil Train

[cx500]

Euclid:  Just take a look through some of the Canadian TSB investigation reports available on their website and you will find a number that involve tank cars.  I am sure the NTSB site will also have them.  It will take some digging to determine which accident reports are relevant but you seem to have the time.  If it is meticulous detail you want, it is certainly there.  Note that these derailments will often involve general trains that include tank cars rather than solid "oil trains".  But the lessons are the same.

Only certain accidents will receive detailed study.  Most times the cause is quickly identified and any corrective action is immediately obvious, if prevention was even reasonably possible.  The detailed investigations will be carried out where the root causes are poorly understood or complex and there may be significant learning opportunities to prevent similar ones in the future.

[Euclid]

Electroliner 1935

One thing I am curious about is have the derailed cars had breeches to their walls or have the appliances (valves, and covers) been the source of the oil spilled in the derailment? I know the 117 cars are supposed to have skids or something to protect the underside valve. Obviously when a car derails, there are excess forces on any and everything. So are the cars splitting at a weld, being punctured or how have they failed? Are they rupturing do to heat created by the fire after the derailment and the fire is from oil that has come out of what opening. This is not clear to me. 

 

One thing I would like to know more about is to what extent mid-train derailments are investigated.  I am referring to derailments not involving collisions or fatalities.  Considering the quest for more crash-proof tank cars, I would expect that these oil train derailments would be investigated in extreme detail.  I would expect them to be measured, photographed, 3D scanned, and modeled in such detail that the entire sequence of events can be played out like a movie, showing the progression of all of the damage and the exact conflicts that caused that damage. 

 

Maybe somebody here knows the extent of investigation and reconstruction that takes place with these oil train derailments, and whether the results are ever made public.  

 

I have known of many derailments, and visited some of them shortly after they happened, but I have never seen even a generalized investigation report of any of them.  But I have seen highly detail reports for collisions in which people were killed or injured.  Those reports go all the way back into the 1800s. 

 

I know that tank cars are vulnerable to puncture, and to valve and fitting damage.  I am not sure about welds splitting.  I would guess that the weld is as strong as the parent material.  It would be interesting to see a full description of the various ways that pileups produce an opening the tank car walls.

 

Punctures are common.  Punctures can initiate a tearing open of the tank wall.  I would expect that deep and sharp bending from wall impact deformation could cause the steel to bend and crack.  I believe that in many cases, the tanks are subjected to extreme compression that causes deep bending, and also sometimes raises the internal pressure high enough to burst the vessel.  I suspect that this pressure explosion may sometimes combine with the deep bending and weakening at the bend line to cause a sudden bursting out of the tank car wall following along a bend line.

 

I would think that loaded tank cars sliding over the ballast would be quickly torn open from the intense abrasion combined with metal wall upset and gouging.  Interestingly however, in the Lynchburg wreck, those eight tank cars ahead of the pileup were dragged on their bottoms and sides for a considerable distance without apparent breaching.  They shed their trucks, and became a chain of cylinders plowing cylindrical furrows through the ballast, while demolishing track ties and rails.

[MidlandMike]

LensCapOn

This is off the current way the thread is flowing, but is a reminder That the danger doesn’t come from the train part of an oil train.

 

“Detroit Highway Closed After Huge Tanker Explosion”

 

http://jalopnik.com/detroit-highway-closed-after-huge-tanker-explosion-1706638360

 

Petroleum sitting in a storage tank is not the problem.  It's when the tank is involved in a truck, train or other transportation accident.  The train wreck is the incident that directly causes the firey train wreck.

Does your car have a gas tank?  Millions of cars go down the road safely every day.  It's usually somebody behind a steering wheel, or road hazard, that set off the bad event.

[zugmann]

Euclid

However, in the U.S., it is considered to be too risky to have that switch under the control of the engineer.

 

We're entrusted with millions of dollars of equipment and enough chemicals to level a large town if we impropely run a train, but we can't be trusted with a switch.

What an industry.

[LensCapOn]

This is off the current way the thread is flowing, but is a reminder That the danger doesn’t come from the train part of an oil train.

 

“Detroit Highway Closed After Huge Tanker Explosion”

 

http://jalopnik.com/detroit-highway-closed-after-huge-tanker-explosion-1706638360

[Euclid]

I have been told the following by a credible source: Nearly all freight cars in the U.S. are equipped with empty/loaded sensors of the mechanical-pneumatic type.  Eliminating the sensors and switching an entire unit train between high and low braking force is a practice used in South Africa ECP, but not in the U.S.  The functionality to do that is built into ECP.  However, in the U.S., it is considered to be too risky to have that switch under the control of the engineer.    

[Euclid]

 

Paul,

You make several good points.  This is my take on them:

ECP might indeed be considered to be a solution in search of a problem.  It offers advantages, but the industry is getting along fine without them, and the universal conversion would have an enormous cost.  However, if ECP is a solution in search of a problem, this newly emerged oil train issue may just be the problem it is searching for.

The point of empty/load sensors is to shorten stopping distance, and it is true that this does not prevent derailments.  However, shortening stopping distance can reduce the pileup effect and the number of cars involved in the derailment.  Thus it can reduce potential damage, injury, and death.  So that is a benefit even if it does not prevent derailments.

Regarding going into emergency versus letting the car drag; either one could cause problems, but there is a third option.  That is to initiate a service application that has less potential to cause the dragging car to jackknife, while still stopping without prolonged dragging.  The service or softer application is not an option with conventional air brakes, but it is with ECP.  ECP could provide the electrical power for derailment sensors whereas conventional air brake systems cannot.  But in the bigger picture, ECP can provide the brake controlling effect of derailment sensors without the sensors.

The electrical connectors have to be reliable.  I assume that this is possible.  Cartier Mining Railway uses ECP in Quebec, so it must work in cold weather.

The up to 70% reduction of stopping distance with ECP compared to conventional air brakes is only for “Service” application of brakes, not for the “Emergency” application.  In a derailment, it is almost always the “Emergency” application that happens, and it happens automatically, by default when the train parts in the derailment process.  The stopping distance advantage of ECP in the “Emergency” application is only 7-14% depending on the braking arrangement of the non-ECP train it is being compared to. 

[Paul of Covington]

ruderunner

Wow I got a convert?

 

   I didn't know if you were referring to me, but I spent some time looking back in this discussion and maybe so.   I've been reading this thread and refrained from commenting for the first 20 pages or so while I digested it.   Some of my thoughts:

    ECP brakes as a solution to the oil train disasters is a solution in search of a problem.   The only justification I remember being advanced was that it MIGHT (or might not) have prevented one incident.

   Empty/loaded sensors on unit trains would not make much difference in preventing derailments, but could be useful in shortening stopping distances.

   Derailment sensors: is it worse to throw a train into emergency if a truck derails or to let it continue dragging it, causing damage to track and switches and leading to more cars derailing?   I would guess the latter.   Any derailment sensor should be simple and sturdy, and not require any source of electrical power.

   About ECP:  I cringe at the thought of 120, 130 or more electrical connections in a train in all extremes of weather.   MC636 says they've been in use successfully in Australia, but I wonder if the area they work in has the extremes of weather we have here.   What is the protection from moisture and dirt on the connectors if a car is set out for a while?

   One last thought on ECP:  stopping distances "up to 70%" shorter.   Ever see a store advertise a sale with prices "up to 70%" off?   Ever find that item?

[ruderunner]

Wow I got a convert?

[Paul of Covington]

   Thanks for your response, Wizlish.   I wasn't thinking in terms of emergency application.

   One more thought and I'll bow out.    My uneducated opinion now is that it might make more sense to expend more effort to improve the empty/loaded sensor to make it more reliable and design it as much as possible to fail in the empty position.    One or two cars with diminished braking in the train shouldn't make much difference.

[Euclid]

Quote by Dave Klepper:

 

Gee Wiz! Who needs all this Buuck Rogers stuff for a modern, purpose-built UNIT OIL TRAIN!

 

All oil cars in the train are loaded or empty, and when loaded are identacle with identacle loads.

 

So one switch for the whole train should be sufficient, controlled from the head end or manually set on each car when loading or unloading. In the latter case the position of the switch must be obvious on the walk-by inspection before the train rolls.

 

Loose car railroading is a problem, but implementation for unit trains shoulo be easy, if the equipment is dedicated.

 

 

Dave,

 

That is the way I look at it.  Apparently load sensors are not preferred on a widespread basis due to their first cost, maintenance expense, and possible failure that may damage wheels.  It may be that ECP offers improved load sensor systems, but the railroads do not prefer ECP either, so better load sensing though ECP has been a non-starter.  But now, oil trains pose a problem that load sensing could remedy, and the ECP brakes are mandated.  So it might be the perfect convergence for oil trains, ECP, and ECP-based load sensing.  ECP brakes reduce stopping distance in emergency applications by 7%.  How much would ECP reduce stopping distance when combined with load sensors?

[Wizlish]

Paul of Covington

I'm not trying to sound snarky, but why do we need a switch to select empty or loaded for the whole train? Wouldn't an engineer know whether he has an empty or loaded unit train and apply the brakes at an appropriate level?

The whole discussion is predicated on the understanding that 'appropriate level' is emergency braking, 'big-holing the Westinghouse', applying the control in the cab as hard as it will go.  If we were talking about modulating the service brake for minimum controlled/safe stopping distance (which is precisely where I thought the control application ought to be made) it would make some sense to keep the engineer's 'hand' in the loop.  Even there, I'd prefer (as with antilock braking of the usual kinds) to have automatics determine the best modulation moment-to-moment until the train has come to a stop.

In emergency, the situation is de facto out of the engineer's hands when he has moved the lever to emergency position (or the trainline has parted and done the same to the brakeline, or the repeater valves in the EOTD or MTDs have dumped the trainline pressure, etc)  Any further adjustments of the brake apparatus will have to be done automatically, and that very particularly applies to sensing and, if necessary, adjusting the braking ratio.  It is also true that any adjustments made to vary braking ratio prior to an emergency application -- which is what Euclid's little two-position switch would do -- can't be modulated further once the train is in an emergency application.

[Paul of Covington]

   I'm not trying to sound snarky, but why do we need a switch to select empty or loaded for the whole train?    Wouldn't an engineer know whether he has an empty or loaded unit train and apply the brakes at an appropriate level?   He or she is not a dumb robot who can't figure things out.

   Sorry, but I'm a bit touchy on this kind of thing, having worked (not railroad related) under management who treated us as if we were incapable of figuring things out for ourselves.

[BaltACD]

daveklepper

Gee Wiz!   Who needs all this Buuck Rogers stuff for a modern, purpose-built UNIT OIL TRAIN!

All oil cars in the train are loaded or empty, and when loaded are identacle with identacle loads.

So one switch for the whole train should be sufficient, controlled from the head end or manually set on each car when loading or unloading.   In the latter case the position of the switch must be obvious on the walk-by inspection before the train rolls.

Loose car railroading is a problem, but implementation for unit trains shoulo be easy, if the equipment is dedicated.

Even in 'unit train' operations, for a variety of reasons, not all loads actually contain full loads of product and not all empties are actually empty.  Any load/empty braking determination must be done on a car by car basis, not by a switch on the locomotive.

[daveklepper]

Gee Wiz!   Who needs all this Buuck Rogers stuff for a modern, purpose-built UNIT OIL TRAIN!

All oil cars in the train are loaded or empty, and when loaded are identacle with identacle loads.

So one switch for the whole train should be sufficient, controlled from the head end or manually set on each car when loading or unloading.   In the latter case the position of the switch must be obvious on the walk-by inspection before the train rolls.

Loose car railroading is a problem, but implementation for unit trains shoulo be easy, if the equipment is dedicated.

 

[Wizlish]

Euclid

But you say there are many solutions, and I am not seeking the best one. What would be the best one?

As we agreed ... and agreed ... and agreed, one which provides proportional braking force based on actual weight, on a car-by-car basis, which provides quick and positive response to wheelslide (again on a car-by-car basis), and which can provide effective differential braking using cars that may have disparate weights.

The two-position system accomplishes none of these, yet requires much of the expensive componentry of an ECP system.  Now it makes sense to use expensive equipment to compress transmission  bandwidth in DTV because you're using the modulation effectively.  With 'two-speed' proportioning, you're just being cheap.

So, now that the FRA has mandated ECP, you have a communication cable that can be used for other things such as sensor data. That cable also opens the door to switching brake force for loaded or empty trains to accomplish what those pesky load sensors do without needing them. It seems like a win-win to me.

I had never quite appreciated the jokes about carbon-fiber buggy-whip shafts or titanium hypersonic yaw strings before now.  You're going to use a 230V line, modulated as a communication bus, to turn your brake response (properly ignoring that for the locomotives and buffer cars) to one of two positions, regardless of actual car weight or car/brake condition?  Once you have the power and data buses, you can use load cells at the center bearings or in the sideframes and get both the average and instantaneous loadings very simply.  For only a slight increase in complexity you can get the load and force data multiplexed with individual car (or truck) ID, so there is no confusion even about which end of a car is experiencing particular forms of vibration.

Meanwhile of course you haven't described how the engineer knows that all the cars have gone to the desired 'loaded' or 'light' state.  Lights on the car frames won't assure this, and do NOT even assume someone is going to check indicators as part of walking the train for a brake test.  You also can't do an on/off light, since you're essentially having to discriminate three states, as with some 'binary' modulation schemes: high, low, and off/disconnected/broken. 

[diningcar]

Way too much ?"?

Let's  just allow the soloist to continue communicating with him/herself

[Euclid]

So just set the cover cars for permanent brake force for a loaded car.  Then when you switch the whole train from loaded to empty, the cover cars remain set for loaded.  Why should that be a big problem?  Or you could just leave the cover cars set for empty, which I assume is where they are set by default.  Maybe they are set to loaded by default.  If so, just leave them set there.

 

My suggestions are based entirely on what I think is the best solution.  I have no stake or agenda that is influencing my suggestions.  Earlier in this thread, I was told that the mechanical-pneumatic load sensors are a maintenance headache, and the can therefore fail.  When they fail, they can flatten wheels or worse.  I presume this is why they are not commonly used. 

 

So, now that the FRA has mandated ECP, you have a communication cable that can be used for other things such as sensor data.  That cable also opens the door to switching brake force for loaded or empty trains to accomplish what those pesky load sensors do without needing them.  It seems like a win-win to me.     

 

But you say there are many solutions, and I am not seeking the best one.  What would be the best one?

[dehusman]

Euclid

Well, for what I am talking about regarding the load/empty unit train brake force selector switch, they will have to stop making small exceptions to uniform load/empty consists.

The cover cars aren't "small exceptions".

There are lots of ways to to have load/empty sensing.  You aren't really trying to find the best solution, you are trying to justify the solution you have chosen.

[Euclid]

Well, for what I am talking about regarding the load/empty unit train brake force selector switch, they will have to stop making small exceptions to uniform load/empty consists.  That would just be the price to pay for the greater good of being able to stop faster.  In the big picture, there has to be exceptions made to accommodate system changes.  Otherwise there will be some detail standing in the way of any significant change in anything.

 

Although there is another way around the problem.  You could develop the control system to have the master switch for the whole train, and then be able to set exclusions to the whole train setting.  The system would have to know where the loads and empties were in the train in order to execute the command to exclude individual cars from the whole train setting.  However, this would leave more room for error.      

[dehusman]

Every "empty" train will have one or two loads (the cover cars).  There is also no guarantee that there won't be loads on an empty train or empties on a loaded train.  If there was a shop at the refinery for the cars, then there might be B/O empties on the loaded train to go to the home shop.  Every once in a while a railroad will move a block of cars on a unit train (such as loads of diesel fuel to an on line fueling location).  You may also have a car that was not loaded but not switched out. 99.99% of the time you are right.   But this whole discussion on oil trains is about exceptions, dealing with the small fraction that has a problem.  If you are dealing with small numbers then small exceptions matter. 

[jeffhergert]

We really don't distinguish trains as being hazmat or non-hazmat for operating restrictions.  We use the term, "key train."  A key train is one that meets certain thresholds of hazmat loads. 

An empty ethanol train is not a key train, although it is considered to have hazmat residue.  A mixed manifest with 20 loads of certain types of hazmat, or lesser number of certain hazmat loads is a key train.  If the mixed manifest only has 19 loads of the lesser nasty stuff, but 10 cars of empty-residue cars, it is not a key train.  Even though it has 29 cars of hazmat/residue hazmat. 

That's the basics, without going into more detail about the certain types of hazmat. 

Jeff  

 

[Euclid]

Can you elaborate?

[BaltACD]

Do hazmat trains consist of loads and empties?

 

Both

[Euclid]

Do hazmat trains consist of loads and empties?

[dehusman]

Euclid

Sensors could be used for that purpose on unit trains, but as you say, all cars in unit trains are either loaded or empty; so the loading of each car is known by the load/empty status of the whole train.

Incorrect in the case of hazmat trains (what we are discussing) and there will be exceptions on other types of trains.

[Euclid]

daveklepper

I presume that load-empty controls are not generally applied to freightcars today.  Unit trains in truth don't need them, because they are normally all full or all empty.  (And thus the buckling argument against ECP does not apply to unit trains.)  And it is unit trains that generate the revenue that would make ECP a wise investment in addition to whatever safety improvements it makes.

Unit trains need empty/loaded brake force adjustment as much as any train, but they don’t need sensors to determine whether a car is loaded or empty.  Sensors could be used for that purpose on unit trains, but as you say, all cars in unit trains are either loaded or empty; so the loading of each car is known by the load/empty status of the whole train.  As discussed earlier, I believe the entire train could be switched to set the brake force on all cars simultaneously from low brake force for an empty train to high brake force for a loaded train.   

 

I am not sure what you mean when you say, “And thus the buckling argument against ECP does not apply to unit trains.”

 

What buckling argument is there against ECP?  ECP reduces the buckling tendency. 

[Euclid]

In the FRA reply to my inquiry, they said this:

“Additionally, ECP brake systems allow for all cars in the train to brake at the same braking (or deceleration) rate even if they had varying physical brake configurations; which is something that cannot be achieved on conventional pneumatic systems. This ability of ECP cars to adjust their effective net braking ratio (NBR) further adds to ECP’s ability to keep run-in forces to a minimum and thus allow the railroad to potentially operate with a higher train-average NBR.”

 

How does EPC do this?  It sounds like he is describing empty/loaded sensing.  He makes it sound like that is a fundamental attribute with ECP, as opposed to an add-on option.  He says it can’t be achieved on conventional pneumatic braking. 

Yet, if he is referring to empty/loaded sensing, it can in fact be achieved on conventional air brake systems as an add-on option.  So he is either wrong, or he is referring to empty/loaded sensing as being a fundamental attribute of ECP rather than an add-on option.

I have never understood empty/loaded sensing to be a fundamental attribute of ECP, but if it is, it would further reduce stopping distance beyond the reduction of stopping distance offered by the instantaneous application of ECP brakes.  

But the reduced stopping distance of ECP is always explained by the instant application.  Other than in the above quote, I have never seen ECP stopping performance explained by optimizing the net braking ratio of each car in the train.  If the NBR were factored in, I would think the stopping distance reduction of ECP would be way higher than 7% for emergency applications, as reported earlier in Wabtec charts.       

[daveklepper]

I presume that load-empty controls are not generally applied to freightcars today.  Unit trains in truth don't need them, because they are normally all full or all empty.  (And thus the buckling argument against ECP does not apply to unit trains.)  And it is unit trains that generate the revenue that would make ECP a wise investment in addition to whatever safety improvements it makes.

[Paul of Covington]

   Euclid:

     "So wheel slide is definitely related to brake force, but it has nothing to do with slack action."

     I agree.   If the brakes are not locked up, which they shouldn't be with load-empty sensing on any type of brake system, it shouldn't matter how much they are pushed by slack action.  I think they have decided to mandate ECP and don't want to be confused by the facts.